It was time to get wet and test the Canon 8 – 15 mm fisheye on the GH5 in the pool so I made my way to Luton Aspire with the help of Rec2Tec Bletchley.
I had the change to try a few things first of all to understand the store coverage of the fisheye frame, this is something I had not tested before but I had built a little model.
This model ignores the corners the red circle are 90 degrees light beams and the amber is the 120 degrees angle. A strobe does not have a sharp fall off when you use diffusers so this model assumes your strobe can keep within 1 Ev loss around 90 degrees and then drop down to – 4 Ev at 120 degrees. I do not want to dig too deep into this topic anyway this is what I expected and this is the frame.
You can see a tiny reflection of the strobes together with a mask falling on the left hand side… In order to test my theory I run this through false colour on my field monitor, at first glance it looks well lit and this is the false colour.
As you can see the strobes drop below 50 at the green colour band and therefore the nominal width of those strobes is probably 100 degrees. In the deep corners you see the drop to 20 % 10% and then 0 %.
Time to take some shots
The lens is absolutely pin sharp across the frame, I was shooting at f/5.6 in the 140 mm glass dome.
Performance remains stunning across the zoom range. I also tried few shots at f/4
There is no reef background but looks pretty good to me.
The pool gives a strong blue cast so the shots are white balanced.
If you want details of the rig and lens mount are in a previous post
The issue is that the Ambient light filters are set for a certain depth and water conditions and does not work well outside that range. While the idea of white balancing the scene and getting colour to penetrate deep into the frame is great the implementation is hard.
Thinking about Keldan we have a 6 meters version and a 12 meters version as listed on their website. The 6 meters version works well between 4 and 12 meters and the other between 10 and 18. At the same time the Spectrum filter for the lens works down to max 15 meters and really performs better shallower than 12 meters.
With that in mind it follows that if you plan to use the spectrum filter -2 you are probably getting the 6 meters ambient filters. So what happens if you go deeper than 12 meters? The ambient light filter is not aligned to the water ambient light and the lights start to look warm this is not such a bad thing but can get bad at times.
You can of course white balance the frame with the lights however this becomes somewhat inconvenient so I wanted to come out with a different technique. In a previous post I have described how to match a lens filter to a light/strobe filter. Instead of matching the light filter to the ambient light I match the filters on land between each other in daylight conditions to obtain a combination that is as much as possible neutral. I have done this for URPRO, Magic Filter and Keldan Spectrum filter and worked out the filter that when combines give a neutral tone.
This tone tends to emulate the depth where the filter has the best color rendition. So in case of Keldan this is around 4 meters and so is Magic with URPRO going deeper around 6-9 meters.
The idea is that you can use the filter without lights for landscape shots and when you put the lights into the mix you can almost shoot in auto white balance or set the white balance to the depth the two were matching. I wanted to try this theory in real life so I did 3 different days of diving testing the combination I had identified the results are in this video.
The theory of matching filters worked and the filter more or less performed all as expected. I did have some additional challenges that I had not foreseen.
The specific performance of a filter is dependant on the camera color science. I have had great results with URPRO combined with Sony cameras but with Panasonic I always had an orange cast in the clips.
Even this time the same issue is confirmed with the URPRO producing this annoying cast that is hard if not impossible to remove also in post.
The Magic filter and the Spectrum filter performed very close, with magic giving a more saturated and baked in image with Keldan maintaining a higher tone accuracy. This is the result of the design of the filters: the Magic filter has been designed to take outstanding picture better than life, the Spectrum filter has been designed using tools to give accurate color rendition. What it means is that the magic images look good even in the LCD while Keldan are a bit dim but can be helped in post.
Looking at the clip in the first 3 and half minutes you can’t tell apart Magic and Spectrum down to 9 meters, with the URPRO giving consistent orange cast.
Going a bit deeper I realised you also need a scenario where you are swimming closer to a reef and want to bring some lights in the frame because you are outside the best working range of the filter. In order to avoid excessive gap when approaching the reef I had stored white balance readings at 6 9 12 15 meters so when I had a scene with mixed light instead of balancing for say 15 meters and then having an issue with the light I used the 9 meters setting so the image is dim when you are far and gets colorful as you approach which is somehow expected in underwater video.
The section at 15 meters are particularly interesting
You can see that URPRO gets better with depth but also how at 5:46 you see a fairly dim reef at 5:52 I switch on the lights and the difference is apparent.
At 6:20 the approach with Keldan was directly with the lights the footage still gives an idea of depth however the colours are there and the background water looks really blu as I had white balance set for 9 meters.
All filters produced acceptable results however I would not recommend URPRO for the Panasonic GH5 and settle for the Magic Filter or the Spectrum filter. Today the spectrum is the only wet filter for the Nauticam WWL-1 but I am waiting for some prototypes from Peter Rowlands for the magic. I would recommend both the magic and the spectrum and the choice really depends on preference. If you want a ready look with the least retouching the magic filter is definitely the way to go as it produces excellent ready to use clips that look good immediately in the LCD.
The Keldan Spectrum filter has a more desaturated look and requires more work in post but has the benefit of a more accurate image.
I think this experiment has proved to work and I will use this method again in the future. This method is also potentially available using the keldan or other ambient light using a tone that closely matches the lens filter.
I think the subject of frame rates for underwater video is filled with a level of non-sense second to none. Part of this is GoPro generated, the GoPro being an action cam started proposing higher frame rates as standard and this triggered a chain reaction where every camera manufacturer that is also in the video space has added double frame rate options to the in codec camera.
This post that no doubt will be controversial will try to demistify the settings and eliminate some fundamental misconception that seem to populate underwater videography.
The history of frame rates
The most common frame rates used today include:
24p – used in the film industry
25p – used in the PAL broadcasting system countries
30p – used in the NTCS broadcasting system countries
PAL (Phase Alternating Line) and NTSC (National Televion System Committee) are broadcasting color systems.
NTSC covers US South America and a number of Asian countries while PAL covers pretty much the rest of the world. This post does not want to in the details of which system is better as those systems are legacy of interlaced television and Cathodic Ray Tubes and therefore are for most something we have to put up with.
Today most of the video produced is consumed online and therefore broadcasting standards are only important if you produce something that will go on Tv or if your footage includes artificial lighting that is connected to the power grid – so LED does not matter here.
So if movies are shot in 24p and this is not changing any time tomorrow why do those systems exist? Clearly if 24p was not adequate this would have changed time ago and except some experiments like ‘The Hobbit’ 24p is totally fine for today use even if this is a legacy of the past.
The human eye has a reaction time of around 25 ms and therefore is not actually able to detect a moving object in the frame at frame rates higher than 40 frames per second, it will however detect if the whole room moves around you like in a shoot out video-game. Our brain does a brilliant job of making up what is missing and can’t really tell any difference between 24/25/30p in normal circumstances. So why do those exist?
The issue has to do with the frequency of the power grid and the first Tv based on Cathodic Ray Tube. As the power of the grid runs at alternate current with a frequency of 60 Hz in the US when you try to watch a movie on Tv that has been shot at 24p this has judder. The reason is that the system works at 60 cycles per second and in order to fit your 24 frames per second there is a technique called Telecine. To make it short artificial fields are added each 4 fields so that this comes up to 60 per second however this looks poor and creates judder.
In the PAL system the grid runs at 50 Hz and therefore 24p movies are accelerated to 25p and this the reason the durations are shorter. The increased pitch in the audio is not noticeable.
Clearly whey you shoot in a television studio with a lot of grid powered lights you need to make sure you don’t have any flicker and this is the reason for the existence of 25p and 30p video frame rates. Your brain can’t tell the difference between 24p/25p/30p but can very easily notice judder and this has to be avoided at all costs.
When using a computer display or a modern LCD or LED Tv you can display any frame rates you want without issues therefore unless you are shooting under grid power artificial lights you do not have to stick to any broadcasting system.
180 Degrees Angle Rule
The name is also coming from a legacy however this rule establishes that once you have set the frame rate your shutter speed has to be double of that. As there is no 1/48 shutter 24/25p are shot at 1/50s and 30p is shot at 1/60s this makes sure also everything stays consistent with possible flicker of grid powered lights.
The 180 degrees angle rule gives each frame an amount of motion blur that is similar to those experienced by our eyes.
It is well explained on the Red website here. If you shoot slower than this rule the frames look blurry if you choose a faster shutter speed you eliminate motion blur so in general everybody follows this and it works perfectly fine.
Double Frame Rates
50p for PAL and 60p for NTSC are double frame rates that are not part of any commercial broadcasting and today are only supported officially for online content.
As discussed previously our reaction time is not able to detect more than 40 frames per second anyway so why bother shooting 50 or 60 frames per second?
There is a common misconception that if you have a lot of action in the frame then you should increase the frame rate but then why when you are watching any movies you don’t feel there is any issue there even if you are watching Iron Man or some sci-fi movie?
That is because those features are shot well with use of a lot of equipment that makes the footage rock steady, the professionals that do it follow all the rules and this looks great.
So the key reason to use 50p or 60p has to do with not following those rules and not being that great of shooting things in a somehow unconventional manner.
For example you hold the camera while you are moving for example a dashboard cam, or you hold the camera while running. In this case the amount of changes in the frame is substantial as you are moving not because things around you are moving. So if you were still in a fixed point it will not feel like there is a lot of movement but if you start driving your car around there is a lot of movement in the frame.
This brings the second issue with frame rates which is panning again I will refer to Red for panning speed explanation.
So if you increase the frame rate from 30 to 60 fps you can double your panning speed without feeling sick.
Underwater Video Considerations
Now that we have covered all basics we need to take into account the reality of underwater videography. Our key facts are:
No panning. Usually except some cases the operator is moving with the aid of fins. Panning would require you to be in a fixed point something you can only do for example in a shark dive in the Bahamas
No grid powered lights – at least for underwater scenes. So unless you include shots with mains powered lights you do not have to stick to a set frame rate
Lack of light and colour – you need all available light you can use
Natural stabilisation – as you are in a water medium your rig if of reasonable size is floating in a fluid and is more stable
The last variable is the amount of action in the scene and the need of slow motions – if required. The majority of underwater scenes are pretty smooth only in some cases, sardine runs, sea lions in a bait ball there really is a lot of motion and in most cases you can increase the shutter speed without the need to double the frame rate.
When I see video shot at 50/60p and played back at half speed for the entire clip is really terrible and you loose the feeling of being in the water so this is something to be avoided at all costs and it looks plain ugly.
Furthermore you are effectively halving the bit rate of your video and to add more usually the higher frame rate of your camera is not better than the normal frame rate of your camera and you can add more frames in post if you wanted to have a more fluid look or perform a slow motion.
I have a Panasonic GH5 and have the luxury of normal frame rates, double frame rates and even a VFR option specifically for slow motions.
I analysed the clips produced by the camera using ffprobe to see how the frames are done and how big they are and discovered a few things:
The 50/60p recording options at 150 Mbps have a very long GOP essentially a full frame is recorded every 24 frames while the 100 Mbps 25/30p records a full frame every 12 frames. So the double frame rate has more frames but is NOT better at managing fast moving scenes and changes in the frame.
The VFR option allows you to set a higher frame rate and then slows down recording to the frame rate of choice. For some reason the 24p format has more options than all the others and the 25p does not even have a 50% option. As the footage is recorded at 100 Mbps the VFR footage at half speed conformed to 30p is higher quality than 60p slowed down to 30p (100 Mbps vs 150/2=75 Mbps) in terms of key frames and ability to predict motion this is better as it has double the amount of key frames per second see this explanation with details of each frame look for the I frames.
The AVCI all intra option has actually only I frames and it will have 24/25/30 of them per second and therefore it is the best option to detect fast movement and changes in the frame. If you need to slow this down this still has 12 key frames per second so other frames can easily be interpolated.
Slow motion – as the image will be on the screen for longer and it is slowed down you need to increase the shutter speed or it will look blurry. So if you intend to take a slow mo you need to make that decision at time of your shot and go for a 90 or 45 degree angle. This remains through if you use VFR or if you slow down AVCI clips in post
If you decided AVCI is not for your the ProRes choice is pretty much identical and again you do not need to shoot 50/60p unless you have specific situations. In general AVCI is equal or better than ProRes so the whole point of getting a recorder is highly questionable but that is another story.
For academic purposes I have compared the 3 different ways Final Cut Pro X does slow down. To my surprise the best method is the ‘Normal Quality’ which also makes sense as there are many full frames.
Now it is interesting to compare my slow motion that is not ideal as I did not increase the shutter speed as the quality of AVCI is high the footage looks totally fine slowed down
Looking at other people example you get exactly the wrong impression you take a shot without increasing the shutter speed and then slow it down. The reason why 60p looks better is for the shutter speed not for the image quality itself it is also completely unneeded to slow down a whale shark as it glides through the water.
So taking this kind of guidance blindfolded is not a good idea.
Key Take Aways
Unless you shoot using main grid powered lights you can choose any frame rate you want 24/25/30 fps.
Shutter speed is important because it can give a motion blur or freeze motion in case of a slow motion clip
You need to choose what scenes are suitable for slow motion at time of capture
Slowing down systematically your footage is unnatural and looks fake
Using formats like AVCI or ProRes gives you better option for slow down than 50/60 fps implementation with very long GOP
VFR options can be very useful for creating purposes although they have limitations (fixed focus)
How do I shoot?
I live in a PAL system country however I find always limitations with the 25 fps options in camera. The GH5 VFR example is not the only one. All my clips are shot 24 fps 1/50s, I do not use slow motion enough and if I did I would probably keep using AVCI and increase the shutter speed depending on the effect I want to give to the scene, this is also the most natural and easier way to shoot underwater as you do not have to continuously change format. Having all intra frames gives me all the creativity I need also for speed ramps that are much more exciting than plain slow motion see this example.
Looking at Nauticam port chart the only option for a fisheye zoom is to combine the Panasonic PZ 14-42 with a fisheye add on lens. This is a solution that is not that popular due to low optical quality.
So micro four thirds users have been left with a prime fisheye lens from Panasonic or Olympus…until now!
Looking at Nauticam port chart we can see that there is an option to use the Speedbooster Metabones adapter and with this you convert your MFT camera to a 1.42x crop allowing you to use Canon EF-M lenses for cropped sensor including the Tokina 10-17mm fisheye. This is certainly an option and can be combined with a Kenko 1.4x teleconverter giving you a range of 14.2 to 33.8 mm in full frame equivalent or 7.1 to 16.9 mm in MFT terms fisheye zoom of which the usable range is 8 -16.9 mm after removing vignetting.
A further issue is that the Speedbooster gives you another stop of light limiting the aperture to f/16 while this is generally a bonus for land shooting in low light underwater we want to use all apertures all the way to f/22 for sunbursts even if this means diffraction problems.
Looking at nauticam port chart an extension ring of 30mm is recommended for the speedbooster and now we have extra 5mm in length Wolfgang uses a 35mm extension. however looking at the lens entrance pupil I have concluded that 30mm will be actually better positioned. Nauticam have confirmed there won’t be performance differences. You need to secure the ring on the dome before final assembly.
The rig looks bigger than the 4.33 dome but the size of the GH5 housing is quite proportionate. It will look bigger on a traditional small size non clam style housing.
The disassembly will be made again in 3 steps.
I am not particularly interested in the 1.4x teleconverter version consider that once zoomed in to 15mm the lens is horizontally narrower than a 12mm native lens so there is no requirement for the teleconverter at all.
This table gives you an idea of the working range compared to a rectilinear lens along the horizontal axis as diagonal is not a fair comparison. The lens is very effective at 8-10mm where any rectilinear would do bad then overlaps with an 8-18mm lens. The choice of lens would be dictated by the need to have or not straight lines. The range from 13mm is particularly useful for sharks and fish that do not come that close.
Horizontal Linear Eq
Wolfgang has provided me with some shots that illustrate how versatile is this set up.
As you can see you can even shoot a robust ghost pipefish!
The contrast of the glass dome is great and the optical quality is excellent. On my GH5 body there is uncorrected chromatic aberration that you can remove in one click. Furthermore lens profiles are available to de-fish images and make them rectilinear should you want to do so.
I would like to thank Wolfgang for being available for questions for providing the 3D print and the images that are featured here on this post.
If you can’t print 3D and need an adapter ring I can sell you one for £7 plus shipping contact me for arrangements.
Note: it is possible to use a Metabones Speed Booster Ultra in combination with a Tokina 10-17mm zoom fisheye and a smaller 4.33″ acrylic dome.
UK Cost of the canon option: £3,076
Uk Cost of the Tokina option: £2,111
However if you add the glass dome back
UK Cost of Tokina with glass dome: £2,615
The gap is £461 and if you go for a Vitrox adapter (would not recommend for the speedbooster) the difference on a comparable basis is £176 which for me does not make sense as the Canon optics are far superior.
So I would say either Tokina in acrylic for the cost conscious or Canon in glass for those looking for the ultimate optical quality.
I was checking the technical details of Alex Mustard Underwater Photography Master Class and the majority of wide angle pictures are taken with a fisheye lens. In the section about shooting sharks Alex says that he prefers to shoot sharks with a fisheye otherwise they look ‘skinny’.
If you look online on underwater video forums you frequently see comments on problems with wide angle lenses connected with the use of a rectilinear wide angle lens in a dome.
The two most common complaints are soft corners and distortion.
Soft corners are due to a combination of lens optical issues and dome port optics. In short any lens is to some extent curved and therefore if you shoot a flat surface the image may be sharp in the centre and softer as you move to the corners. Issues with field of curvature are corrected stopping down the lens. The issue with field of curvature happens everywhere not just underwater.
Right now there are four wide angle lens that can be housed for a micro four third camera:
This lens has a nice working range that allows to capture 100 degrees diagonal at widest setting and still has a 35mm equivalent at the tele end. This is a pretty little lens at $699 is the most affordable option that can be put in a housing. You will need a wide angle port and the zoom gear. The whole combination for your Nauticam housing comes at $1,399. This lens can also be combined with a glass dome but this will make the whole combination much more expensive and you may want to think about getting a better lens instead.
This is an outstanding lens especially on land due to the fast f/2.8 aperture. It is expensive at $1,299.99 and very heavy and bulky. The lens does not fit through the N85 port opening and requires a port adapter this gives the extra benefit of a focus know but with such a wide lens is not really useful due to high depth of field. You will need a 180mm glass dome and the zoom gear for the lens to complete the set up ending at a whopping $3159.99.
I have owned this lens and I have to say that at $799 is the right compromise between wide field of view and price. Furthermore once you get the zoom gear you have the option of a cost effective acrylic dome that will give you a very wide set up for $1589.99. There are reports of poor performance with this lens and it is true that is not as sharp in corners but the results are perfectly acceptable if you stop at f/8 in close shots.
This lens is prone to reflections and flare however once you add the N120 port adapter and the 180mm glass dome this will get you to $2819 at that point you may want to consider the Olympus combination instead.
This is my favourite lens is sharp does not suffer from field of curvature issues and has a very useful zoom range 16-35mm in 35mm equivalent. The zoom gear and the 7″ acrylic dome will take you to 1889.99 that is an excellent price point. The lens is not prone to reflection or flare and as the 7″ dome has the same curvature radius than the 180mm dome it will produce very similar results.
The significant size of the acrylic port and the fact it floats make it ideal for split shots and this is the lens that gives me the best results.
This lens can also take port adapter that allows you to use the 180mm glass dome. This adds up to $2919.99 if you experience bad reflections and shoot frequently in the sun it may be worth it but I have not had any issue so far with this lens probably because of its nano coating.
I have found the 7mm focal length too problematic for dome ports and the amount of perspective distortion excessive generally it would be preferred to shoot at 9mm and narrower however this maybe insufficient for wreck interiors if you want a rectilinear look.
One of the regular complaints of video shooters especially in wrecks or caves is that the edges look horrible and distorted and that there is an issue with the corners pulling. This is in fact not an issue but a problem with perspective as you shoot very wide angle. The following test shots will illustrate that the issue happens on land and has nothing to do with dome ports.
As we can see the football looks like an oval and the chair is pulled. This is due to a perspective issue and is not a lens problem. When you shoot underwater video the objects on the edges of the frame change shape creating this pull effect that most people dislike.
At 9mm the amount of perspective distortion is reduced and this is the reason why 18mm on 35mm equivalent is one of the favourite focal length for rectilinear video and the maximum angle that should be used in small spaces to avoid the pulling edges.
One of the reason why a lens like the Nauticam WWL-1 is preferred for video is because the corners look sharp but is that really true?
Not really let’s apply some barrel distortion to simulate the WWL-1 to the image that looked badly distorted.
Now the football looks circular as we have applied -60 barrel distortion, obviously the rest of the image is now bent but this seems not to be of a concern to most people!
It needs much less correction to bring the 9mm shot into shape and for sure between the 8mm and 9mm the 9mm is the dimension that produces the most acceptable results.
It has to be said that in video with 16:9 aspect ratio most of the issue will be cropped away at the edges but the distortion in the middle of the frame will remain. For the same reason the 9mm image will appear practically rectilinear with no issues
I hope this post was useful there are four options for micro four thirds shooters to use rectilinear lenses I have settled for the Panasonic 8-18mm as in most cases it is still possible to control the perspective issue, I found this impossible at 7mm.
Obviously if you shoot in the blue this problem will not be visible however rectilinear lenses are popular with wreck shooters and I think this posts gives an idea of the challenges at play.
Finally I would discourage the use of the 7-8mm focal length range for video to those that want to have a rectilinear look.
From this post I started supporting Bluewater Photo in US for my links because it still provides multi brand and choice and because I learnt a lot from Scott Gietler Underwater photography guide back in the days where there was no internet resource to learn from.
It is possible to buy a larger card such as the reference that is 7.5″x10″ however this is cumbersome and I found the Studio version to work well with the Panasonic GH5 as it only uses the central part of the frame for white balance.
Going back to our GH5 instruction manual you can also see that the camera white balance is limited to 10,000K which is the colour of blue sky.
Underwater due to light absorption at longer wavelengths red and orange disappear at depth and blue tends to scatter over suspended particles. So the colour temperature of water tends to be higher than 10,000K and also the blue is somewhat washed out by scattering.
This is the reason filters are essential because reduce the amount of blue or to say better cyan and bring the camera into a range where custom white balance works again.
I have already posted a whole range of observations on filters in a previous post so am not repeating here.
With the right filter for the water colour I dive in and with the appropriate white balance card you can get some pretty decent results with custom white balance.
To help the colour accuracy I have experimented with the Leeming Luts and I want to thank Paul Leeming for answering my obscure questions. Obviously you do not have to use the LUTs and you can design them yourself however I found that using the Cinelike D LUT I have a very good starting point for colour correction.
The starting point is a CineLike D profile with saturation, noise reduction and sharpness set to -5 all other settings to default as suggested by Paul, there is no need to lower the contrast as CineLike D is already a flat curve.
*Noise and sharpness have actually nothing to do with grading but are set to -5 as the GH5 applies sharpening and noise reduction even at -5 setting. Sharpening has generally a negative effect all around while noise reduction if required is better performed in the editor.
Looking at imaging resource tests of the GH5 we can appreciate that the camera colours are oversaturated by default.
The GH5 tends to push deep colour and wash out cyan and yellow. This becomes apparent when we look at a white balanced clip uncorrected.
The Leeming Lut helps rebalancing the camera distorted colours and when you apply the camera LUT, provided you have followed the exposure instructions and applied the profile as described, the improvement is immediate.
From here onwards it is possible to perform a better grading and work to improve the footage further.
One other thing that I believe it is interesting is that while generally for ambient light or balanced light shots I do not actually trust the camera exposure and go -1/3 to -2/3 for close up shots exposing to the right greatly helps highlights recovery
In the two frames you can see the difference the LUT brings restoring the correct balance to the head of the turtle.
To be clear the turtle detail has been white balanced in water on the whibal card while using a Keldan Spectrum filter -2, then in fcpx automatic balancing is applied. The LUT brings out a better dynamic range from the same frames.
Obviously you are free to avoid lens filters and LUTs and to some extent it is possible to get similar results however the quality I obtain using automatic settings I believe is quite impressive.
I found myself most times correcting my own wrong exposures or wanting to increase contrast in scene where I had little however this only happens in sever circumstances where white balance and filters are at the limits.
There are many paths to get the right colours for your GH5 underwater videos in my opinion there are four essential ingredients to make your life easier and give your footage a jump start:
Take a custom white balance using a professional grade white balance card
Set the right picture profile and exposure when shooting
(Recommended) Use appropriate filters for the water conditions
Apply the appropriate LUT to eliminate the errors in the GH5 colour rendering in post processing
With the following settings producing a video like this is very simple and all your efforts are in the actual cutting of the clip.
Please note some of the scenes that look off are shot beyond the working conditions of filters and white balance at around 25 meters…
I see many posts on line debating which macro lens is best for your micro four third system.
If I refer to the Nauticam system we have 4 macro lenses:
For the purpose of this article I will skip the Olympus 30mm as the Panasonic lens is known to be sharper and will focus on the other 3 lenses.
DxOMark is a popular tool for comparison as it gives you the results on one page. I have run it for the Oly 60 and the Pana 30 and 45 on the 20 Mpix OMD E-M1 MKII
Surprisingly the much more expensive Leica performs worse than the other cheaper models, this is confirmed on all internet sites running other type of tests.
What we can see is that there is little difference between the Panasonic 30mm and Olympus 60mm when it comes to image quality so whichever lens you choose your subject at the same level of magnification and aperture will have more or less the same detail.
Common Misconception: Shorter focal length give more depth of field
Many people think that using a longer lens is harder because there is less depth of field this is actually incorrect conceptually.
Enter for the Panasonic GH5 the following parameters
Circle of confusion: 0.015 mm
Focal length 30mm
Distance 10.5 cm (minimum distance of the 30mm Macro)
Result Total depth of field 0.3cm
Focal Length 60mm
Distance 21 cm (as it achieves the same magnification)
Result Total depth of field 0.3cm
So depth of field is not a consideration when choosing a macro lens…
Shooting a subject close to the background
In the following 3 shots am taking an image of a widget at f/11 at 29-44-60 mm on a Leica 12-60 (it is just easier it makes no difference to the outcomes)
At the same aperture you can clearly see that there are no difference whatsoever in the detail and actually overall in the picture you don’t notice anything.
Shooting a subject far from the background
For the second shot I have moved the widget away from the wall.
Again there is no difference in the level of detail of the widget however looking at the background we can see that
The 60mm shot reveals one brick and less than one quarter
The 44mm shot reveals one brick and a half
The 29mm shot reveals two bricks
So while the subject is exactly the same as the 60mm lens has a narrower field of view we see much less of the background.
This means that if you are shooting a nudibranch on the sand or something flat on a rock you won’t notice anything however if there is space behind the subject you will capture much more of that resulting in less subject isolation.
Underwater Comparison 30 vs 60 mm
In the first shot the Rhinopia is taken with a 60mm lens
In the second shot the same Rhinopia with the 30mm (in a different place to be fair)
I have marked up with red the areas that with a longer focal length would have been minimised.
Which Lens to choose?
Now that we have clarified that depth of field is not a consideration and as each macro lens will have the same magnification there are only two factors that matter:
Isolation from background
The 60 mm will have a longer working distance and at the same magnification will isolate the subject better from busy backgrounds, the 60 mm is also better for skittish subject because of the longer working distance. I have this lens and I have borrowed the 30mm in couple of occasions but do not have the 30mm yet.
There are however situations where longer working distance is not a benefit, specifically when the visibility is poor and there are suspended particles or the subject is really large.
In the example below I was at one meter from the two frogfish, a 30mm would have been better however the shot came OK.
I believe the Olympus 60mm is a must lens to have. To date I have not felt the need for the Panasonic 30mm that is indeed a very sharp lens because I have always managed to pull out the shots. However for someone diving in murky water and focussing on nudibranches or subject laying flat on the seabed the Panasonic 30mm could be a better choice. I also want to say that using the 14-42mm at 42mm for me is actually a better choice for portraits and with a close up lens works very well with small subject not super macro and therefore as I already own the 14-42mm and various diopter for me the 30mm is not on the shopping list.
There is no doubt that the Panasonic GH5 is a very capable camera and in given conditions the video performance you can get is truly impressive.
Broadly speaking a video clip needs to be:
Those 4 characteristics are tightly related to:
Today everyone shoots 4K and after all resolution is well supported by almost any camera, broadly is unaffected by other factors and unless the noise is really high sharpness of your frame is not going to be a real issue shooting at 4K.
In normal conditions and not underwater a camera can resolve many colors. However underwater due to the diffraction of light and selective absorption of colours the starting point is very different from land. So generally is not the camera that cannot resolve the colors but the colors that are missing to start with. This post will focus specifically on this aspect. The Panasonic GH5 can resolve 23.8 bits in RAW and therefore technically has less than 8 bits color depth – do not confuse this with the 8 or 10 bit recording setting.
Underwater scenes tend to have limited dynamic range, with the exception of sunbursts or shooting against the light this is going to be an issue only in specific circumstances of very bright scenes with shadows. In all scenes taken with video lights dynamic range is not an issue at all. The GH5 has 13 stops of dynamic range but rarely this is an important consideration.
Noise is an important consideration as when the noise goes up the camera looses the other characteristics, color, dynamic range and resolution will be affected when the camera is outside the sweet spot. Broadly speaking the Panasonic GH5 does not do well once you pass the ISO 1600 setting and I tend to cap the ISO in video at 800 in most cases.
To understand how those variables play we can see how the same set up reacts very differently in scene where there is less light and therefore the camera uses high ISO like this one.
The same camera with exactly the same equipment in brighter water produces this
So the reason for the above is that with less light there are less colours and the clip looks what it is really.
OK moving on to the main subject of this post how do I get the colors right? It is a combination of techniques and the trick is to use the right one in the right conditions.
Generally every site has specific conditions that change depending on weather, time of the day, visibility and other factors. So in broad terms a site will have more or less light and therefore more or less colours. It is therefore impossible to categorically define what to do at a given depth but is more about typical values. With this in mind we have typically 3 scenarios:
Ambient light shots
Artificial light shots
Balanced light shots
Close up Shots
In general close up shots especially of small subject fall within the scenario 2 for which a video light with high color rendering is important as this will define the colours you see. With a lot of power it is possible to extend artificial lighting to larger subjects but eventually you run out of power due to distance or size of the subject.
Wide angle shots and seascapes
True wide angle shots are generally ambient light shots which also means when it gets too dark the colors will be missing and it will look blue not matter the equipment.
In order to make the most of ambient light shots for wide angle it is essential to balance the colours in water even when you use a RAW format on a still image because RAW files are not as RAW as you think and are actually compressed.
Custom White Balance
Using Custom White balance with a grey card it is possible to obtain decent results until the camera hits the maximum color temperature in the case of the Panasonic GH5 this is 9900K. Depending on conditions you may get to 10-12 meters and this still works, in darker water this stops working much sooner.
Color filters push the limit of custom white balance further down. Some add more or less 4 meters others up to 6-8 meters at the expense of an overall loss of light. Filters are useful when there is a lot of light because also help to keep the Panasonic Gh5 in the best aperture range (not smaller than f/11)
Right now there are predominantly 3 filters on the market:
All those filters will improve the performance and color rendering of your footage, under the conditions that the loss of light is not pushing the camera above reasonable ISO values.
In terms of depth range the magic filter and the Keldan Spectrum -2 version can be pushed to 15 meters depth on a bright day in clear water. The URPRO is capable of getting a few meters more down to around 17-18 meters although it does generate an orange cast (as there is no red left) it is still workable.
Typical Max Depth
1 2/3 Ev
Keldan Spectrum -2
2 stop (WWL)
1 2/3 Ev
This image gives an idea of the 3 filters as you can see they are very different one from the other.
Balanced Wide Angle Shots
This is an entirely new technique that has started with the Keldan Ambient light filters. I wrote a whole piece on wetpixel
The principle is to use custom white balance with or without filter to obtain color rendering and then put filters on the video light so that the color of the light emulates the ambient light and therefore it only gives texture not color.
Keldan has developed a whole range of filters for various situation that match their light and therefore are not applicable to any other light.
As I do not own a set of Keldan I have done some tests and found that a gel of Cyan filter 2 or 3 stops makes my divepro G18+ practically ambient light in the conditions I dive into.
The above value are based on my experience use at your own risk especially with different lights.
To give an idea I overlapped the filter to my iPhone lens
This is the shot without any filters
This example shows that the two filters cancel themselves the result is almost daylight with no cast which means in water if you use a video light or a strobe you will not see a red or orange spots on the image.
For those taking pictures the same combination remains true with Inon Z240 and Sea and Sea YS-D2
Example picture here
One thing to take into account is that you need to find a way to hold the gel on the video light or the strobes. The flat surface strobe diffusers make this process easy, finding something you can use with your video lights is not easy and also the gels may melt after continuous use.
It comes a point and a depth where filters stop working, this could be as shallow as 8 meters in green water. As the scene is dark using lights is what is required. There is nothing specific about this technique except making sure you don’t get burned highlights or backscatter. As it happens in photography using long arms (maybe not as long as for stills) is key to get good lighting on your subject.
My Camera Settings
I use CineLike D with saturation, sharpness and noise reduction to -5. I shoot at 24/25p AVCI 400 mbps and follow the 180 rules, it is entirely possible to shoot at 1/100 if you like more crisp look.
Clearly there are people out there that do not like filters and think white balance is best etc but I think a good read on magic filters explains it all.
It has been almost 4 years since my first review of the Nauticam WWL-1 wet wide angle lens and a few accessories later this lens is definitely my all time favourite for underwater video with my GH5.
I do not want to repeat myself and beat to death the topic of sharpness in corners I would rather recap on the other benefits of this lens that really make it unique for underwater video. Obviously this lens is very valid also for still images because of the ability to zoom through but this is not the focus of this post.
So let’s have a look at the three killer features of this lens that make it really special
Field of view
The WWL-1 once combined with the Panasonic 14-42mm MKII (the best lens to combine with the WWL-1 in my view) offers a field of view of 130 degrees diagonal. But what does that really mean?
First the WWL-1 does not compare with a rectilinear lens in fact it is almost a fisheye lens as we can see from those shots of a pool wall.
The barrel distortion is evident correcting the image in lightroom gives an idea although not 100% correct of what is the real field of view of the lens.
What is interesting to see is that the WWL-1 like a fisheye lens offers a much wider diagonal field of view than on the other dimensions.
I have compared the WWL-1 with other rectilinear lenses and with the 8mm fisheye.
Linear Ratio to FE
The table I have prepared uses the equisolid equation for a fisheye lens to map the WWL-1 I have verified the values and I can confirm the WWL-1 is somehow equivalent to 10.06mm fisheye lens.
There are two things that are worth noting, the first is that on the horizontal and vertical axis the WWL-1 is not wider than the Panasonic 7-14mm at 7mm. The other consideration is that with the WWL-1 the 4:3 format frame starts to become a classic 3:2 as the ration width/height is 1.56.
When we work in video at 16:9 we crop out most of the diagonal part leaving the rest of the field of view intact this means that in video mode the lens is much more rectilinear and the barrel distortion more contained.
If we look at a frame at 25mm we can see that at 4:3 the level of distortion is reduced but still present.
.The corrected frame shows the residual distortion.
The level of residual distortion in video mode is pretty negligible at 25mm. At 35mm even in 4:3 mode the WWL-1 is practically straight.
The benefit of the distortion of the WWL-1 is such that if you are shooting large sharks for example the barrel distortion makes those sharks look large in the centre of the frame and when they go out of the frame you don’t have the pull effect of a classic rectilinear lens behind a dome. At the same time if you need to shoot some divers or lines that are straight you can zoom in and still cover a pretty wide field of view.
The other benefit of the WWL-1 is that allows you to use lenses that are stabilised, today any lens at the 7-8mm range on micro four third has no stabilisation which means you need to use the in body stabiliser if available with your camera. For the GH5 this means that choosing a lens like the 14-42 MkII gives you access to Dual IS combining body and lens stabiliser and IS lock that really is useful when finning around. I just want to make sure that I am clear I am talking of this lens
The Panasonic 14-42PZ power zoom is NOT compatible with dual IS so if you use this lens you either have in body or lens stabiliser not both. Obviously if you have a camera without stabilisation like the GH5s or the BMPCC 4K all of this is less relevant but still you can have some stabilisation instead of nothing.
I have not shot a comparative Dual IS vs Lens IS vs nothing in the pool but I am planning to do that soon. I can only say once you have dual IS with IS lock you don’t want to go back.
Personally I think that any dive down to 18 meters in tropical or subtropical water will benefit from a filter but I also believe that conditions may change and in some cases you want to take the filter off. Now most of the rectilinear lenses for the GH5 do not even take a filter but also consider that once you fit one in dome port you are stuck with it for the dive. With the WWL-1 and the Keldan filter if you feel there is too little light and you want to get rid of the filter you can.
I managed to get hold of a pair of DivePro G18 Plus samples last week to use for my wide angle video hereby my thoughts on this product that I believe could be interesting for many video shooters.
Construction and technical data: solid anodised aluminium smoothly finished. Jaunt has decided to expose the battery that screws into the light head as oppose to house the battery itself into the light a more conventional design. The light features a Cree CXB3590 LED with a color temperature of 5000K and a CRI of 92 with declared output of 18,000 Lumens and a 95 degree beam angle. The battery delivers 14.4V with a total capacity of 6800 mAh giving an autonomy of 52 minutes at full power of 105 minutes at half power. The technical specifications of Creed LED give full details of the light used https://www.cree.com/led-components/media/documents/ds-CXB3590.pdf
When you look at page 14 performance group EB you see 15000 lumens nominal per bin, take into account that the array has an efficiency of 120% so this gives you the nominal lumens. In terms of viewing angle the nominal beam in air is 115 degrees and this in water gives you 100 degrees according to my calculations am not sure how Jaunt worked out 95 degrees. Looking at page 10 you can see that even at CRI=90 there is a spike in the navy blue colour this is most likely to show in water when combined with close up lenses as blue fringing and should be completely irrelevant at wide angle.
The provided chargers delivery only 1A with a declared charging time of 8 hours that frankly is totally unacceptable for a video light.
I recommend getting a spare battery the code is DivePro B06. The light has a color indication for the charge level around the switch button that is simple and effective at the same time although you don’t get your residual time in minutes as you do from some other products.
Ergonomics: the light feels very much like a torch and when attached to the included ball mount is very bottom heavy. The fresh water weight is 350 grams and becomes 370 grams with the mount. The light feels very well built, smooth and rugged.
The light switch system allows for 100% 50% and any intensity from 1% to full power however this is slightly difficult to use, there is also an SOS feature. Generally the light feels well designed with few small niggles, the ball attachment is quite long this is not an issue except the light is already heavy on the bottom whilst the attachment to the arm is in the front part this creates a significant torque especially on land. Use of a bespoke underwater float is advised. Another minor issue of the ball mount is that it is not actually 1” but 2.65 cm. 1mm may seem little but it does bring the clamp off balance making it easier to loose grip. I recommend changing the o-rings with normal buna rings to reduce the size and improve grip if you use Ultralight, Inon or Nauticam arm systems. Another small issue of the switch system is that at rest on land the button may hit the floor. Previous version of the light switch on immediately at button press now Jaunt has changed the logic of the switch on to a long 2 second press to avoid accidentally switching on the light.
As the light overheats outside water this will prevent LED burnout. I had the latest version of the light, other copies on the market may still have the old logic check carefully yours.
Field test: Testing the light in a tank confirms the beam angle and the battery life as per specification more detailed testing is only possible in the pool.
I tested the lights in a pool with a 5 meters deep end. Pools have controlled conditions to check how the lights fall off and what is the real life coverage of the lights for your lens. It is not possible to provide a real test of the geometry of a light in the sea unless you know exactly the dimensions of what you are shooting and you have flat surfaces. I have some diving end of April in the Mediterranean will provide an update later on open water performance especially in terms of color rendering that I have not checked in the pool.
Beam angle: I took a series of stills at 80, 100, 150 from the axis of the lights using arms 28” long to which you need to add the length of the tray to determine where the light beams will meet.
As expected at 80 cm you see two separate circles of light, this become an elliptical shape at 100 cm and fill the frame at 125 cm with small fall off at the edges.
This pool shot gives an idea of the coverage only the edges are dark and the light is very nicely distributed. I have removed the reflections of the lights on the wall as they are distracting.
In order to understand the light falloff I ran the stills into a monitor equipped with false colour you can see the results here the lights are in my opinion very convincing and still deliver at 1.5 meters distance although you may need to pump up the ISO.
I would think that with this angle of coverage a distance of 30” between the two lights is ideal with my rig this means two 8” segments however this may create ergonomic issues so I settled at 5 + 8 this gives an arm length of 21” and with the GH5 26” from centre. Considering a WWL-1 set up with the front of the lens 6” ahead this means shooting distances up to 1 meters from the front of the lens covering a frame width of around 3.5×2.0 m that is a huge surface.
After the pool test I noticed a number of scratches in the paint clearly the coating is not the same quality of more expensive lights.
Overall the lights have great performance in terms of power, beam distribution and are very well built with excellent autonomy. Issues to note and suggested remediation:
The chargers that come with the lights are inadequate. A 2A charger would be better and would halve the charge time without damaging the battery pack, here some links for third party 16.8V/2A chargers:
The clamp position creates an issue of pitch in water obviously if you manage to have some floatation system for the individual light this is no longer a problem.
The light intensity adjustment is not effective and you can’t really tell if the two lights are set the same. Obviously this is only an issue if you need less than 9000 lumens for example for close shots. There is no easy solution to this problem other than controlling the two lights simultaneously.
Considering the price point and the quality this light competes with the Gates GT14 and the Keldan 8X and when it comes to power, CRI, autonomy and price beats them both in terms of durability and ergonomics they appear to be a level down but they are also less expensive.
Jaunt is setting up relationships with distributors in major markets and in UK there is already one so everybody should be able to get a set of lights. Depending on location, duties and import regulations the price will change but generally I have seen it remain competitive.